Deep Learning Analysis of Smart Meter Data from a Small Sample of Room Air Conditioners Facilitates Routine Assessment of their Operational Efficiency

This is the official code repository for the paper Deep Learning Analysis of Smart Meter Data from a Small Sample of Room Air Conditioners Facilitates Routine Assessment of their Operational Efficiency, which is published in Energy and AI.

1. Environment Setup

All experiments were conducted under Linux CentOS system with Anaconda (Python 3.9) as the developing environment.

Use the following pip command to install all the required packages:

pip install -r requirements.txt

2. Data Compilation

Due to the privacy issues, the dataset will not be made open to public.

However, we provide a 200 lines sample version of the full dataset to demonstrate the formation of our experimenting data, and you can check the preprocessing/data_compilation.py for how our data is compiled from different categories of data.

Remarks: Please notice that the Location in sample_data.csv are set to 0 for privacy.

3. Model Training & Evaluation

All codes for Setting I are stored in setting_1 directory, and Setting II are stored in setting_2 directory. Training codes are in their respective folders, and scripts used for training are stored in setting_1/script and setting_2/script.

We also provide the evaluation codes and visualized results as shown below.

Setting I overall performances:

Setting II overall performances:

Regarding the performance of different models, we also made the following visualization plots for comparisons.

Comparison between five models in Setting I.

Comparison between five models in Setting II.

4. Energy Saving Result

After applying our best model on unlabelled rooms, we acquire the total electricity energy saving results by comparing the electricity energy consumption distribution between normal RACs with poorly efficiency RACs. We verify our models on the data collected in 2022/2023, and the results are shown below.

5. Acknowledgement

This project was supported by the Undergraduate Research Opportunity Program (UROP) of The Hong Kong University of Science and Technology (HKUST) and the Sustainable Smart Campus project of HKUST. The authors would also like to thank the anonymous reviewers for their valuable comments and suggestions. The views and ideas expressed here belong solely to the authors and not to the funding agencies.

6. Citing This Work

Please use the bibtex below for citing our work

@article{WANG2024100338,
    title = {Deep Learning Analysis of Smart Meter Data from a Small Sample of Room Air Conditioners Facilitates Routine Assessment of their Operational Efficiency},
    journal = {Energy and AI},
    pages = {100338},
    year = {2024},
    issn = {2666-5468},
    doi = {https://doi.org/10.1016/j.egyai.2024.100338},
    url = {https://www.sciencedirect.com/science/article/pii/S2666546824000041},
    author = {Weiqi Wang and Zixuan Zhou and Sybil Derrible and Yangqiu Song and Zhongming Lu}
}

7. Contact

If you have any question, feel free to email me at mightyweaver829 [at] gmail.com. This email will be active all the time.